JP2014176033A - Communication system, communication method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically generate minutes with accuracy.SOLUTION: A communication system including a plurality of information processing devices connected through a network includes: first voice conversion means for converting voice information acquired by one of the plurality of information processing devices into character data; in another information processing device among the plurality of information processing devices, second voice conversion means for converting into character data the voice acquired according to the voice information acquisition timing by the one information processing device; and determination means for comparing two character data converted by the first and second voice conversion means to determine whether or not the voice information acquired by the one information processing device and the other information processing device is the same speech.

Description

  The present invention relates to a communication system, a communication method, and a program.

  A remote conference using an information processing apparatus between a plurality of bases connected to a network is known. In such a communication conference system, an invention has been proposed in which the speech content of a conference is automatically converted into text and recorded by automatic speech recognition. Patent Document 1 proposes an invention in which minutes are easily created using automatically generated text information. In other words, in Patent Document 1, for the purpose of easily creating the minutes of a meeting, the comments posted in the minutes, such as important statements, are detected as index information and inserted into the minutes file simultaneously with the progress of the meeting. Discloses a configuration of a video conference system that can automatically create a simple minutes.

  In automatic speech recognition, it is technically difficult to recognize speech with 100% accuracy, and generally 60 to 90% of the entire text information can be correctly recognized in a situation without special noise. Therefore, the text information contains many errors.

  Therefore, it is necessary to correct this text in order to create an accurate minutes, but since the original error rate is high, a correction error occurs in the process of correcting the text information, and there is information that differs from the actual content. There is a possibility that it will be listed in the minutes.

  Therefore, Patent Document 2 proposes an invention that registers a user dictionary in order to increase the certainty of voice recognition and makes it easy to detect frequently used words.

  However, in Patent Documents 1 and 2, although the accuracy of automatic speech recognition is improved and the creation of minutes is improved, the error rate is still high, and the technology is further improved in order to automatically generate the minutes correctly. There was a need.

  In view of the above problems, an object of the present invention is to provide a communication system, a communication method, and a program capable of automatically generating the minutes more accurately and automatically.

In order to solve the above problems, according to one aspect of the present invention,
A communication system having a plurality of information processing devices connected via a network,
First voice conversion means for converting voice information acquired by one information processing apparatus of the plurality of information processing apparatuses into character data;
Second voice conversion means for converting voice information acquired according to the timing at which voice information was acquired by the one information processing apparatus into character data in another information processing apparatus among the plurality of information processing apparatuses. When,
The two character data converted by the first and second voice conversion means are compared, and it is determined whether or not the voice information acquired by the one information processing apparatus and the other information processing apparatus is the same utterance. Determination means to perform,
A communication system is provided.

  According to the present invention, the minutes can be automatically generated more accurately.

1 is an overall configuration diagram of a communication conference system according to an embodiment. FIG. The functional block diagram of the information processing apparatus which concerns on one Embodiment. The audio | voice information and text data example which concern on one Embodiment. The flowchart which showed the statement record production | generation process which concerns on one Embodiment. The example of the content of the statement of each base chosen as the comparison object concerning one embodiment. An output processing example according to an embodiment. An output processing example according to an embodiment. The flowchart which showed the statement recording production | generation process which concerns on the modification 1 of one Embodiment. The output process example which concerns on the modification 1 of one Embodiment. The flowchart which showed the utterance record production | generation process which concerns on the modification 2 of one Embodiment. The output process example which concerns on the modification 2 of one Embodiment. The flowchart which showed the utterance record production | generation process which concerns on the modification 3 of one Embodiment. The flowchart which showed the statement recording production | generation process which concerns on the modification 4 of one Embodiment. The function block diagram of the communication conference server which concerns on one Embodiment. The hardware block diagram of the information processing apparatus which concerns on one Embodiment.

  DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, preferred embodiments of the invention will be described with reference to the accompanying drawings. In addition, in this specification and drawing, about the substantially same structure, the duplicate description is abbreviate | omitted by attaching | subjecting the same code | symbol.

<Introduction>
A remote conference using an information processing apparatus between a plurality of bases connected to a network is known. In such a communication conference system, it has been practiced to automatically convert the content of a conference utterance into text by automatic speech recognition and record it as a minutes. However, in automatic speech recognition, it is technically difficult to recognize speech with 100% accuracy, and the text information contains many errors.

In particular, the communication environment varies depending on the location where the information processing apparatus is located as follows. For example,
(1) The sound quality varies depending on the type, performance, and arrangement of microphones.
(2) While the voice of the local site is directly input from the microphone, the voice of the other base is input as digitized data via the server. In many cases, the data is encoded.

  Due to such a difference in environment, the result of converting voice information into text may differ depending on the information processing apparatus at each site.

  Therefore, in the communication system of the present embodiment described below, speech information is converted into text information by speech recognition in information processing apparatuses at two or more bases, and text data (hereinafter, (Also referred to as character data) for each utterance, and a technique for identifying whether or not the same utterance is identified. Thereby, the location of the conversion mistake at the time of converting audio | voice information into text data can be reduced. As a result, it is possible to create a highly accurate minutes.

  In addition, when creating minutes, it is very laborious for a person to merge a plurality of text data created by the information processing apparatus at each site. Therefore, in the system of the present embodiment, a plurality of text data can be compared to identify whether they are the same utterance, and the plurality of text data can be automatically merged appropriately based on the identification result. In this way, it is possible to create minutes with less effort by automatically determining whether or not the same statement is made.

  Furthermore, voice information is converted into text information by voice recognition in information processing apparatuses at two or more bases. For this reason, depending on the network conditions, even if the communication performance between certain bases deteriorates and the voice is interrupted, it is possible to reliably make text from the voice information recognized by the information processing device at one of the two or more bases. Data can be obtained.

  The communication system of this embodiment that exhibits the above functions and effects will be described below. In the present embodiment, a communication conference system will be described as an example, but the communication system according to the present embodiment is not limited to the communication conference system. For example, the communication system according to the present embodiment records the exchange of voice information transmitted or received using two or more information processing devices as text data in an interactive information providing system, an interactive window system, or the like. It can be used as a communication system.

[System overall configuration]
First, a communication conference system according to an embodiment of the present invention will be described with reference to FIG. FIG. 1 is an overall configuration diagram of a communication conference system according to an embodiment. The communication conference system 1 according to the present embodiment includes a plurality of information processing apparatuses 10a, 10b, and 10c (hereinafter collectively referred to as information processing apparatus 10) and a communication conference server 50 via an IP network 110. Connected to transmit or receive audio information or the like. The communication conference system 1 according to the present embodiment can support not only audio information but also image / video information, and can function as a video conference system.

  The information processing apparatuses 10a, 10b, and 10c are examples of terminals for communication conferences, and may be PC terminals or tablet terminals. A function of connecting to the IP network 110 is provided at each site. In FIG. 1, the information processing apparatus 10 a is placed at the base A, the information processing apparatus 10 b is placed at the base B, and the information processing apparatus 10 c is placed at the base C. The information processing apparatuses 10a, 10b, and 10c exchange voices via the IP network 110 to establish a conference by communication between remote points.

  The information processing apparatus 10 includes voice conversion means. The voice conversion means 19 is not only at one place in the communication conference system 1, but there are a plurality of voice conversion means 19 in the communication conference system 1. In other words, each information processing apparatus 10 does not necessarily have the voice conversion means 19. There may be a plurality of voice conversion means 19 in the communication conference system 1. For example, in FIG. 1, the information processing apparatuses 10 a, 10 b, and 10 c each have the voice conversion means 19, but the present invention is not limited thereto, and for example, the information processing apparatuses 10 a and 10 b each have the voice conversion means 19, There may be a case where the information processing apparatus 10 c does not include the voice conversion means 19. However, since a plurality of voice conversion means 19 are required in the communication conference system 1, for example, the information processing apparatus 10a has the voice conversion means 19 and the information processing apparatuses 10b and 10c do not have the voice conversion means 19. No way.

  The communication conference server 50 is a device that relays the audio information of the information processing devices 10a, 10b, and 10c arranged at each base. In a video conference system, it is generally called a “multipoint control unit” (MCU). The communication conference server 50 may be configured by software or hardware.

  Note that in the communication conference system 1 according to the present embodiment, communication is performed between the information processing apparatuses 10 arranged at each site via the communication conference server 50. However, the communication conference system 1 according to the present embodiment is not limited to this, and the information processing apparatuses 10 at each site may communicate directly with each other via the IP network 110 without using the communication conference server 50. In the present embodiment, various devices are connected by the IP network 110, but may be connected by other means.

[Functional configuration of information processing device]
Next, the functional configuration of the information processing apparatus according to the present embodiment will be described with reference to FIG. FIG. 2 is a diagram illustrating a functional configuration of the information processing apparatus according to the present embodiment. The information processing apparatus 10 includes a communication unit 11, a data processing unit 12, a voice input unit 13, an input voice processing unit 14, a voice storage unit 15, an output voice processing unit 16, a time measuring unit 17, a voice output unit 18, and a voice conversion unit 19. Voice recognition result storage means 20, speech record generation means 21, speech record output means 22, determination means 23, and speaker identification means 24.

The communication unit 11 is a unit that is connected to the IP network 110 and communicates with the other information processing apparatus 10 and the communication conference server 50. The communication unit 11 transmits and receives various digital data including audio information. Specifically, as information to be transmitted / received, the following information is transmitted / received in addition to information necessary for connecting to the IP network 110.
・ Voice information ・ Speech record information ・ Time information ・ Device setting information Basically, the information transmitted from the own device is information generated by the information processing device 10 at the own site, and the transmitted information is transmitted to other sites. In many cases, the information is generated by the information processing apparatus 10. In the communication conference system 1 according to the present embodiment, text data corresponding to voice information needs to be collected in one place after the conference ends. In this case, the final text data is transmitted and received using the communication means 11. The

  The data processing means 12 is a means for processing received information or a means for processing information to be transmitted. The received information and the information to be transmitted are transmitted to the data processing means 12 as acquired information. It is generally considered that audio information is encoded and transmitted / received. In this case, the data processing means 12 encodes transmission information (Encode) and decodes reception information (Decode).

  The voice input means 13 is a means for inputting voice. As an example of the voice input unit 13, a microphone is generally used in the communication conference system 1. The voice input means 13 may include a terminal for inputting data such as a recorder.

  The input voice processing means 14 is a means for processing the voice input from the voice input means 13. The input sound processing means 14 is an A / D (analog / digital) conversion means for converting analog data of sound input from the microphone into digital data, an equalizer for changing / adjusting the frequency characteristics of the sound, and noise of the input data of the sound It is comprised from the noise removal means etc. which remove | eliminate.

The voice storage means 15 is a means for storing voice information. A volatile memory such as a RAM, a non-volatile memory such as an HDD (Hard Disk Drive), and various FLASH memories can be considered. One or more of these may be provided. Examples of the voice information stored in the voice storage unit 15 include the following data.
Input voice digital data processed by the input voice processing means 14 and data obtained by encoding it.
Received audio digital data and decoded data.

  The output sound processing means 16 includes a D / A (digital / analog) conversion means for converting sound digital data into analog data, an equalizer for changing / adjusting the frequency characteristics of sound, an amplifier for amplifying sound, and the like.

  The time measuring means 17 is a means for measuring time, and may be composed of, for example, a clock or a timer. It is desirable that the time measuring means 17 has a function of synchronizing with a clock built in the information processing apparatus 10 at another base. The clocking means 17 counts the transmission time of the voice information, the reception time of the voice information, and the generation time of the voice information as the speech time of the voice information.

  The sound output means 18 is a means for outputting sound, and may be constituted by various speakers and earphones, for example.

  The voice conversion means 19 is means for converting the content of a statement into text data by a voice recognition technique (a technique in which a voice spoken by a person is analyzed by a computer and converted into text data (character data)). Various methods are known as a speech recognition technique used by the speech conversion means 19. In automatic speech recognition, it is technically difficult to recognize speech with 100% accuracy, and generally 60 to 90% of the entire text information can be correctly recognized in a situation without special noise. Therefore, many errors are included in the textual information. For this reason, the information converted into character data includes many errors. By the way, Japanese with many homonyms is known to have a low speech recognition success rate.

  The communication conference system 1 according to the present embodiment requires two or more voice conversion means 19. For example, the voice conversion means 19 that converts voice information acquired by one information processing apparatus among a plurality of information processing apparatuses 10 into character data is defined as the first voice conversion means. The voice conversion means 19 that converts voice information acquired by another information processing apparatus among the plurality of information processing apparatuses 10 into character data is referred to as a second voice conversion means. In that case, the communication conference system 1 according to the present embodiment requires at least two or more voice conversion means 19 of the first and second voice conversion means. For example, the voice conversion unit included in the information processing apparatus 10a at the site A in FIG. 1 may be used as the first voice conversion unit, and the voice conversion unit included in the information processing apparatus 10b at the site B in FIG. 1 may be used as the second voice conversion unit. Good. The voice conversion means included in the information processing apparatus 10b at the site B may be used as the first voice conversion means, and the voice conversion means included in the information processing apparatus 10c at the site C may be used as the second voice conversion means. However, it goes without saying that the system configuration of the present embodiment is an example, and the system configuration is not limited to this as long as it has two or more voice conversion means.

  The voice recognition result storage means 20 is means for storing information obtained by converting voice information into character data using the voice recognition technique in the voice conversion means 19. The voice recognition result storage means 20 also receives and stores information obtained by converting voice information into character data in the information processing apparatus at another base. The voice recognition result storage unit 20 may be shared with the voice storage unit 15.

The utterance record generating means 21 is a means for creating a utterance record using, for example, the following information.
Text data of the information processing apparatus 10 at the local site stored in the voice recognition result storage unit 20
Text data of the information processing apparatus 10 at another base stored in the voice recognition result storage unit 20
Time information timed by the time measuring means 17
-Speaker (speaker) information specified by the speaker specifying means 24, and information that associates a speaker with a speaker.

  The utterance record generating unit 21 generates a utterance record made by communication between one information processing apparatus among the information processing apparatuses 10a, 10b, and 10c and another information processing apparatus based on determination by the determination unit 23 described later. To do.

  The voice recognition result storage means 20 includes means for associating voice data with time, and creates a simple statement record by using the time information of data obtained by characterizing voice information of the information processing apparatuses at the local site and the other sites. The voice recognition result storage unit 20 may create a statement record using only data obtained by converting the voice information of the information processing apparatus at its own site into text. The utterance record data created here is recorded in the utterance record output means 22 or transmitted or output from the communication means 11 to an external device.

The utterance record output means 22 is means for outputting the utterance record data created by the utterance record generation means 21. Various methods can be considered as the output method by the utterance record output means 22. The statement record output means 22 includes at least one output method described below.
-Output as image data (analog RGB component signal, etc.).

In this case, the utterance record output means 22 outputs information to a projector, a monitor or the like. If the message record is updated and displayed in real time, the conference can proceed while confirming the message record. In this case, the utterance record output means 22 requires means for converting the utterance record data into image data.
・ Output the recorded speech data as digital data.

  In this case, specifically, the utterance record output means 22 includes various memories IF such as USB-Host IF and SD Card IF, and various IFs such as RS-232C IF, which have the above functions.

  The determination unit 23 compares two or more character data converted by the two or more speech conversion units 19 (first and second speech conversion units), and determines whether the speech information is the same speech.

  The speaker specifying means 24 specifies the speaker of each voice information by voiceprint authentication.

[Speech record generation processing]
Next, the statement record generation process according to the present embodiment will be described with reference to FIG. FIG. 4 is a flowchart showing a statement record generation process according to the present embodiment. In the description of the flowchart of FIG. 4, in a communication conference between information processing apparatuses at two bases, an example of processing actual text data when voice information is converted into text data by the information processing apparatuses at each base (FIG. 3). Is used.

  FIG. 3A shows the exchange of voice information actually performed in the communication conference. It is assumed that Mr. Sato is at the site A, Mr. Suzuki and Mr. Tanaka are at the site B, and a communication conference is performed by using a total of three information processing apparatuses 10a and 10b. The case where the conversation as shown here is made will be considered below. FIG. 3A shows information about the base 30, the speaker 32, and the content of the speech 34 in the order in which the sound is produced.

  FIG. 3B shows an example of data converted into text by the information processing apparatus 10a at the site A. The text data of the base A in FIG. 3B includes a speech time 36, a base 30, a speaker 32, and a speech content 34. No. of FIG. Tanaka-san's remark “Yes 2” cannot be converted into text data in the information processing apparatus 10a at the base A and is not recorded. Note that the speech time is the time recognized by the information processing apparatus 10 at each site.

  FIG. 3C shows an example of data converted into text by the information processing apparatus 10b at the site B. The text data of the base A in FIG. 3C includes a speech time 36, a base 30, a speaker 32, and a speech content 34. No. of FIG. It is assumed that the word “specification” by Mr. Sato in 5 is misrecognized as the word “use”.

  Based on the above assumptions, the statement record generation processing of this embodiment will be described with reference to the flowchart of FIG. 4 and the state of the text data of FIGS.

  S101: A minutes creation command is notified to the information processing apparatus 10. It is considered appropriate at the end of the meeting, but the timing is arbitrary. The instruction notification method may be any method. For example, a method of providing a minutes creation button in the information processing apparatus 10 and notifying the information processing apparatus 10 of a minutes creation command by pressing the button can be considered.

  S102: The created text data is transmitted to the information processing apparatus 10 that creates the minutes. The minutes may be created by each information processing apparatus 10 or by only one or a plurality of predetermined information processing apparatuses 10. When receiving the minutes creation command, the information processing apparatus 10 transmits text data generated from the voice information to the information processing apparatus 10 that creates the minutes. The information processing apparatus 10 that creates the minutes receives text data transmitted from each apparatus. Note that the apparatus for creating the minutes may be a communication conference server instead of the information processing apparatus 10.

  S103: The text data of each base is arranged in time series (see FIG. 5). Here, it is assumed that the communication between the base A and the base B takes 4 seconds. Therefore, in order to recognize the message of α11 at the base B after 4 seconds, the message times 36 of α11 and β12 are shifted by 4 seconds. In other words, the timing at which one information processing apparatus 10 acquires an α11 message and the timing at which another information processing apparatus 10 acquires a β12 message are within the recent time that is at the same time or about several seconds apart. The acquisition timings of two statements are similar (corresponding). Assume that the text data of each base is arranged in the order of the speech time 36 as shown in FIG.

  S104: The determination means 23 selects the remarks of each base that performs the comparison process. Here, α11 and β12 in FIG. 5 are first compared.

S105: The determination means 23 performs a process of decomposing each utterance into word levels. As an example, α11 and β12 are decomposed as follows. "This is the base / A's / Sato /. Can you hear me?"
S106: The determination means 23 compares the words of each utterance and records the matching words and their appearance order. The cases of α11 and β12 are exactly the same.

  S107: The determination unit 23 determines whether the number of matching words is equal to or greater than a specified ratio. If the number of matching words is greater than the specified ratio, the process proceeds to S108. If the number of matching words is less than the specified ratio, the process proceeds to S111. The “specified ratio” here may be arbitrarily determined. If it can be set as a parameter, it is possible to adjust the level at which it is determined that there is a match. For example, it can be set such that more than 30% of all words match.

  S108: The determination unit 23 determines whether the appearance order of the matched words matches. If they match, the process proceeds to S109. If not, the process proceeds to S111.

If they say the same, the order of appearance should match. Conversely, if the appearance order does not match, even if there are many matching words, they should not be the same statement. Therefore,
S109: When the appearance order of the matched words matches, the determination unit 23 determines that the comparison target is “same statement”.

  S110: In this case, the utterance record generating means 21 performs processing as the same utterance. Specifically, in the example of FIG. 5, α11 and β12 are determined to be the same statement, and the statement record generation unit 21 performs text data of the base A and text data of the base B as α21 and β21 of FIG. The process described in the same line is performed.

  S111: When the appearance order of the matched words does not match, the determination unit 23 determines that the comparison target is “not the same statement”.

  S112: In this case, since the utterances are not the same, the utterance record generating means 21 performs processing described in different lines in the example of FIG. Therefore, there is no text data of base A corresponding to the text data of β22 of base B.

  S113: The determination means 23 determines whether comparison of all the statements has been completed. If it is determined that the process has been completed, the process proceeds to S114. If it is determined that the process has not been completed, the process returns to S104.

  S114: The statement record output means 22 performs an output process and ends this process. If it is unnecessary, nothing may be done and the process may be terminated.

  As an example of output processing by the utterance record output means 22, when base information is important, it is possible to output text data of each base side by side as shown in FIG. Alternatively, when the base information is unnecessary, one piece of text data of the same message may be output without duplication as shown in FIG.

  With the above flow, text data can be recorded accurately and easily. The minutes of 100% complete statement records are not guaranteed, but it is very easy to understand when creating the minutes by referring to the generated record records, and it is easy and accurate in a short period of time. Minutes can be created.

  As described above, according to the communication conference system 1 according to the present embodiment, voice information is converted into text information by voice recognition in the information processing apparatuses 10 at at least two bases. Therefore, each information processing apparatus 10 can compensate for the location of each other's conversion error, and correction errors can be further reduced, and a statement record can be efficiently created as a minutes. Furthermore, a plurality of text data can be appropriately merged by providing a technique for discriminating whether or not the speech is the same. As a result, the minutes creation time can be reduced.

Hereinafter, Modifications 1 to 4 of the statement record generation process according to the present embodiment will be described with reference to FIGS. 8 to 13.
(Modification 1)
FIG. 8 is a flowchart showing a utterance record generation process according to the first modification of the present embodiment, and FIG. 9 is an output process example according to the first modification of the present embodiment.

  Modification 1 is different only in that S110 of FIG. 4 in the above embodiment is changed to S115 of FIG. Specifically, in the above embodiment, if the determination unit 23 determines that the same statement is made, the same statement is included in the statement records of the bases A and B of the one information processing apparatus 10a and the other information processing apparatus 10b. (FIG. 6). On the other hand, in the first modification, when the determination unit 23 determines that the same statement is made, the same statement is included in the statement record on the base side of one of the information processing device 10a and the other information processing device 10b. However, the same utterance is not included in the utterance record of the other base (FIG. 9).

As described above, in Modification 1, when there are remarks determined to be the same and there is no difference in the text data identified by the information processing apparatus 10 between the bases, the text data on one base side is left, and the other The process of deleting the text data on the base side is executed. According to this, since duplicate text data in the minutes in which the utterance is recorded is deleted, there is an effect that the content of the utterance becomes easy to see. In FIG. 9, the text data on the site A side is left and the text data on the site B side is deleted. However, the present invention is not limited to this, the text data on the site B side is left and the text data on the site A side is deleted. Also good.
(Modification 2)
FIG. 10 is a flowchart showing a utterance record generation process according to the second modification of the present embodiment, and FIG. 11 is an output process example according to the second modification of the present embodiment.

  The modification 2 is different only in that S115 of FIG. 8 of the modification 1 is changed to S116 of FIG. Specifically, in the first modification, when the determination unit 23 determines that the same statement is made, the same statement is included in the statement record of one of the information processing device 10a and the other information processing device 10b. The same remark is not included in the remark record of the other base (FIG. 9). In the modified example 2, in addition to this, when it is determined that the same utterance is determined by the determination unit 23 and there is a difference in the same utterance, the utterance record is generated so that the difference portion in the same utterance can be recognized. . For example, in FIG. 11, when the utterances are determined to be the same and there is a partial difference in the text data, the difference is marked. Thus, according to the second modification, there is an effect of making it easy to see the difference included in the remarks determined to be the same between the bases.

In order to display the different parts in the same statement in a recognizable manner, in addition to marking the different parts, make the different parts bold, change the color of the different parts, bracket the different parts, and underline the different parts. Various methods such as those shown in FIG.
(Modification 3)
FIG. 12 is a flowchart showing a statement record generation process according to the third modification of the present embodiment. Modification 3 includes all the processes of FIG. 10 of Modification 2 and further includes step S117 of FIG. Specifically, in the third modification, in S117, which is a step after S104, the determination unit 23 determines whether or not the difference between the utterance times of the selected bases is equal to or greater than a predetermined time. When it is determined that the difference between the speech times of the respective bases is less than a predetermined time, the processes after S105 are executed. On the other hand, if it is determined that the difference between the speech times of the selected bases is equal to or greater than a predetermined time, the process returns to S104 without performing the comparison processing of the selected speech, and then the comparison processing Select remarks to be made.

As described above, in the third modification, the determination unit 23 determines that the difference in the speech time of the speech information before conversion of two or more character data at each base to be compared exceeds a predetermined threshold (predetermined time). In this case, the determination as to whether or not the voice information is the same speech for the two or more character data is stopped. In this way, even when the comparison target is the same, control is performed such that “when the speech time is significantly different, the comparison target is not included”. Thereby, the whole process can be reduced and the processing time can be shortened.
(Modification 4)
FIG. 13 is a flowchart showing a statement record generation process according to Modification 4 of the present embodiment. The modification 4 includes all the processes of FIG. 10 of the modification 2, and further includes the step of S118 of FIG. Specifically, in the modified example 4, in S118, which is a step after S104, the determination unit 23 determines whether or not the selected speakers are different. When it is determined that the speaker at each base is the same person, the processing after S105 is executed. On the other hand, when it is determined that the speaker at each base is different, the process returns to S104 without performing the comparison process of the selected message, and the message for the next comparison process is selected.

  In the information processing apparatus 10 having a voiceprint authentication function, it is possible to acquire information in which a speech is associated with a speaker. In that case, the determination means 23 may compare the speaker 32 first, as shown to S118. For example, both α11 and β12 in FIG. 5 are comparison targets because the speaker is A1, but if these speakers are different, they are not compared. In this way, even if the comparison target is the same, control is performed to stop the determination as to whether or not the voice information is the same speech when the speakers of the speech content to be compared are different. Thereby, the whole process can be reduced and the processing time can be shortened.

  As mentioned above, according to the said embodiment and the modification 1-the modification 4, when producing an utterance record (minutes) based on the text data produced by voice recognition, the audio | voice acquired by two or more information processing apparatuses, respectively. Compare two or more character data converted from information. Thereby, the speech of audio | voice information can be determined correctly. As a result, the accuracy of automatic creation of minutes can be increased, correction errors can be reduced, the minutes of speech can be efficiently created, and the time for creating minutes can be reduced.

  In addition, the said form does not limit the scope of the present invention, and a communication conference server may be provided with a part or all of the determination function of the information processing apparatus and the statement record generation function. In addition, a plurality of communication conference servers and information processing apparatuses that constitute the system may be provided, and either the communication conference server or the information processing apparatus may be provided with the above function. It should be noted that the system configuration in which the information processing apparatus and the communication conference server described in this embodiment are connected is an example, and it goes without saying that there are various system configuration examples depending on applications and purposes.

  As another example of the system configuration, in the functional configuration of the information processing apparatus 10 illustrated in FIG. 2, the voice conversion unit 19, the speech recognition result storage unit 20, the speech record generation unit 21, the speech record output unit 22, and the determination unit 23 and a system configuration that the server 50 has the functions of the speaker specifying means 24 instead of the information processing apparatus 10 may be used. In this case, as shown in FIG. 14, the communication conference server 50 includes the communication means 57, the data processing means 58, the time measuring means 59, the voice conversion means 51, the voice recognition result storage means 52, the statement record generation means 53, It has functions of a utterance record output means 54, a determination means 55, and a speaker identification means 56.

  The functions of the voice conversion unit 51, the voice recognition result storage unit 52, the utterance record generation unit 53, the utterance record output unit 54, the determination unit 55, and the speaker identification unit 56 are the same as those of the information processing apparatus 10. For example, the voice conversion unit 51 converts voice information acquired by one information processing apparatus among the plurality of information processing apparatuses 10 into one character data. The voice conversion means 51 is voice information acquired by another information processing apparatus, and the voice information acquired according to the timing when the voice information is acquired by the one information processing apparatus is converted into other character data. Convert.

  The determination unit 23 compares the one character data converted by the voice conversion unit 19 with the other character data, and the voice information acquired by the one information processing apparatus and the other information processing apparatus. Is the same statement.

  In this system configuration example, the communication conference server 50 executes the statement record generation processing of the above embodiment and each modification. In this case, the statement record generation process shown in FIGS. 4, 8, 10, 12, and 13 is executed on the communication conference server 50 side. At that time, the voice conversion means 51 of the communication conference server 50 converts the voice information acquired by the information processing apparatuses 10 in at least two bases into text information using a voice recognition technique. Therefore, also in the present system configuration example, it is possible to compensate for a conversion error portion of the voice information acquired by each information processing apparatus 10, and thereby it is possible to create a more accurate minutes. The communication conference server 50 corresponds to a server device connected to a plurality of information processing apparatuses 10 via a network.

  The preferred embodiments of the communication system and communication method of the present invention have been described in detail above with reference to the accompanying drawings, but the technical scope of the communication system and communication method of the present invention is not limited to such examples. It is obvious that a person having ordinary knowledge in the technical field of the present invention can come up with various changes or modifications within the scope of the technical idea described in the claims. Of course, it belongs to the technical scope of the communication system and communication method of the present invention. In addition, when there are a plurality of the above-described embodiments and modifications, they can be combined within a consistent range.

  A hardware configuration example of the information processing apparatus and the communication conference server according to the present invention will be briefly described with reference to FIG. The information processing apparatus and communication conference server according to the invention have a CPU 106 built therein. A program for realizing each function executed by the CPU 106 may be stored in advance in a storage unit such as the ROM 104, the RAM 105, or the HDD 108. The program may be recorded on a recording medium such as a CD-ROM or a non-volatile recording medium (memory) such as a flexible disk, SRAM, EEPROM, or memory card. The functions of the information processing apparatus and the communication conference server according to the present invention can be realized by causing the CPU 106 to execute the programs recorded in these memories. Furthermore, the program can be downloaded from an external device connected to the IP network 110 using the function of the communication circuit 103 and having a recording medium recording the program or an external device storing the program in a storage means. The keyboard 101 is an example of an input device, and is used to input each operation signal to each device. A mouse or touch panel can be used instead of the keyboard 101. The display 102 is an example of a display device, and displays a processing result by each device.

  As described above, the information processing apparatus and the communication conference server according to the present embodiment can realize the various functions described above by the hardware configuration.

  1: communication conference system, 10a, 10b, 10c, 10: information processing apparatus, 11: communication means, 12: data processing means, 13: voice input means, 14: input voice processing means, 15: voice storage means, 16: Output voice processing means, 17: timing means, 18: voice output means, 19: voice conversion means, 20: voice recognition result storage means, 21: utterance record generation means, 22: utterance record output means, 23: determination means, 24 : Speaker identification means, 50: Teleconference server, 110: IP network

JP 2005-341015 A JP 2011-066532 A

Claims (11)

A communication system having a plurality of information processing devices connected via a network,
First voice conversion means for converting voice information acquired by one information processing apparatus of the plurality of information processing apparatuses into character data;
Second voice conversion means for converting voice information acquired according to the timing at which voice information was acquired by the one information processing apparatus into character data in another information processing apparatus among the plurality of information processing apparatuses. When,
The two character data converted by the first and second voice conversion means are compared, and it is determined whether or not the voice information acquired by the one information processing apparatus and the other information processing apparatus is the same utterance. Determination means to perform,
A communication system comprising:   2. The statement record generation means for generating a statement record generated by communication between the one information processing apparatus and the other information processing apparatus based on the determination by the determination means. Communication system.   The statement record generation means includes the same statement in the statement records at the respective bases of the one information processing apparatus and the other information processing apparatus when the determination means determines that the same statement is the same. The communication system according to claim 2.   The statement record generation means includes the same statement in the statement record at one of the bases of the one information processing apparatus and the other information processing apparatus when the determination means determines that the same comment is present, The communication system according to claim 2, wherein the same utterance is not included in the utterance record of any other base.   The said utterance record generation means generates the said utterance record so that a difference portion in the same utterance can be recognized when there is a difference in the same utterance when the determination means determines that the same utterance is the same. The communication system according to claim 3 or 4, characterized by the above. Further comprising a time measuring means for measuring the speech time of the voice information;
When the difference between the speech times of the speech information before the conversion of the two or more character data to be compared exceeds a predetermined threshold, the determination means has the same speech information for the two or more character data. The communication system according to any one of claims 1 to 5, wherein the determination as to whether or not is stopped. A speaker specifying means for specifying a speaker of the voice information;
The determination means stops determining whether the voice information is the same speech for the two or more character data when speakers of the voice information before conversion of the two or more character data to be compared are different. The communication system according to any one of claims 1 to 6. A communication system having a plurality of information processing devices and server devices connected via a network,
The voice information acquired by one information processing device of the plurality of information processing devices is converted into one character data, and the one information is converted by the other information processing device of the plurality of information processing devices. Voice conversion means for converting the voice information acquired in accordance with the timing at which the voice information is acquired by the processing device into other character data;
Comparing one character data converted by the voice converting means with another character data, it is determined whether the voice information acquired by the one information processing device and the other information processing device is the same utterance. Determination means to perform,
A communication system comprising: The determination means includes
Each of the voice information acquired by the one information processing apparatus and the other information processing apparatus is decomposed into one or a plurality of words and compared for each word, and the one information processing apparatus and the other information processing apparatus The communication system according to claim 1, wherein it is determined whether or not the voice information acquired in step 1 is the same statement. A communication method performed by a plurality of information processing apparatuses connected via a network,
A first voice conversion step of converting voice information acquired by one information processing apparatus of the plurality of information processing apparatuses into character data;
A second voice conversion step of converting voice information acquired according to the timing at which voice information is acquired by the one information processing apparatus into character data in another information processing apparatus among the plurality of information processing apparatuses. When,
The two character data converted in the first and second voice conversion steps are compared, and whether or not the voice information acquired by the one information processing apparatus and the other information processing apparatus is the same utterance. A determination step for determining;
A communication method characterized by comprising: A program for executing a function of a communication system having a plurality of information processing devices connected via a network,
A first voice conversion process for converting voice information acquired by one information processing apparatus of the plurality of information processing apparatuses into character data;
Second voice conversion processing for converting voice information acquired according to the timing at which voice information is acquired by the one information processing apparatus into character data in another information processing apparatus among the plurality of information processing apparatuses. When,
The two character data converted by the first and second voice conversion processes are compared, and whether or not the voice information acquired by the one information processing apparatus and the other information processing apparatus is the same utterance. A determination process for determining;
A program for running a computer.

Images